1 #!/usr/bin/env python

     2 

     3 import os

     4 import sys

     5 import shutil

     6 import htmlentitydefs

     7 

     8 def replace_entities(line):

     9     i = 0

    10     outstr = line

    11     while True:

    12         estart = outstr.find("&", i)

    13         eend = outstr.find(";", estart)

    14         if estart == -1 or eend == -1:

    15             break;

    16         # entity replace is needed

    17         entitystr = outstr[estart + 1:eend]

    18         i = eend

    19         if entitystr in htmlentitydefs.name2codepoint:

    20             replacestr = str(htmlentitydefs.name2codepoint[entitystr])

    21             outstr = outstr[:estart] + "&#" + \

    22                      replacestr + ";" + \

    23                      outstr[eend + 1:]

    24             i = len(outstr[:estart]) + len(replacestr) + 3

    25                     

    26     return outstr

    27 

    28 infile = sys.argv[1]

    29 for line in open(infile, "r"):

    30 	print replace_entities(line),